Pedestrian violations of traffic signal are common. How to reduce or even eliminate such violations has aroused a heated discussion among researchers and the government. Based on the research on the motive of pedestrian dangerous crossings, we have developed a dynamic adjustment algorithm for signal lights to reduce unsafe crossing behaviors from the perspective of signal setting, which determines when the pedestrian green time starts and how long it takes. We simulated the process of pedestrians becoming impatient as they waited. The expected waiting time curve of pedestrians is u-shaped. Therefore, the method aims to reduce the proportion of pedestrians who reach the expected waiting time threshold, leading to reducing the violation rate of pedestrians. Our algorithms substantially outperform the traditional fixed controller as it perceives the traffic flow (pedestrian arrive rate and vehicle flow) more comprehensively and assigns the green time resource more efficiently than the traditional controller, particularly in the situation of small vehicle volume and high pedestrian arrival rate. Results indicate that the algorithm can significantly reduce the pedestrian violation ratio and improve the performance of the entire road network on the basis of less impact on the vehicle volume. This finding demonstrates that incorporating the pedestrian arrival rate into the consideration of signal design from the perspective of pedestrians and vehicles provides a feasible solution to dangerous pedestrian behaviors.

CICTP